# V-CAGE: Vision-Closed-Loop Agentic Generation Engine for Robotic Manipulation

Source: [V-CAGE: Vision-Closed-Loop Agentic Generation Engine for Robotic Manipulation](<https://www.sota2.com/research/paper/69dc78637d0172140ec7892b>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-14T05:00:04.234Z

Authors: Yaru Liu, Ao-bo Wang, Nanyang Ye

Published: 2026-04-10T06:56:17.000Z

## Abstract

Scaling Vision-Language-Action (VLA) models requires massive datasets that are both semantically coherent and physically feasible. However, existing scene generation methods often lack context-awareness, making it difficult to synthesize high-fidelity environments embedded with rich semantic information, frequently resulting in unreachable target positions that cause tasks to fail prematurely. We present V-CAGE (Vision-Closed-loop Agentic Generation Engine), an agentic framework for autonomous robotic data synthesis. Unlike traditional scripted pipelines, V-CAGE operates as an embodied agentic system, leveraging foundation models to bridge high-level semantic reasoning with low-level physical interaction. Specifically, we introduce Inpainting-Guided Scene Construction to systematically arrange context-aware layouts, ensuring that the generated scenes are both semantically structured and kinematically reachable. To ensure trajectory correctness, we integrate functional metadata with a Vision-Language Model based closed-loop verification mechanism, acting as a visual critic to rigorously filter out silent failures and sever the error propagation chain. Finally, to overcome the storage bottleneck of massive video datasets, we implement a perceptually-driven compression algorithm that achieves over 90\\% filesize reduction without compromising downstream VLA training efficacy. By centralizing semantic layout planning and visual self-verification, V-CAGE automates the end-to-end pipeline, enabling the highly scalable synthesis of diverse, high-quality robotic manipulation datasets.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Long-horizon Robotic Manipulation on AutoCheckout Synthesized](<https://www.sota2.com/research/sota/long-horizon-robotic-manipulation-on-autocheckout-synthesized.md>): Long-horizon robotic manipulation; AutoCheckout Synthesized; Success Rate: not recorded.

- [Long-horizon Robotic Manipulation on PackBreads Synthesized](<https://www.sota2.com/research/sota/long-horizon-robotic-manipulation-on-packbreads-synthesized.md>): Long-horizon robotic manipulation; PackBreads Synthesized; Success Rate: not recorded.

- [Long-horizon Robotic Manipulation on PackStationery Synthesized](<https://www.sota2.com/research/sota/long-horizon-robotic-manipulation-on-packstationery-synthesized.md>): Long-horizon robotic manipulation; PackStationery Synthesized; Success Rate: not recorded.

- [Long-horizon Robotic Manipulation on SortToCabinet Synthesized](<https://www.sota2.com/research/sota/long-horizon-robotic-manipulation-on-sorttocabinet-synthesized.md>): Long-horizon robotic manipulation; SortToCabinet Synthesized; Success Rate: not recorded.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2604.09036>)

[PDF](<https://arxiv.org/pdf/2604.09036.pdf>)
